Você está na página 1de 28

Excavations at Utica by the Tunisian-British Utica Project 2012

Elizabeth Fentress Faouzi Ghozzi Josephine Quinn Andrew Wilson

Excavations at Utica by the Tunisian-British Utica Project 2012


Elizabeth Fentress, Faouzi Ghozzi, Josephine Quinn, and Andrew Wilson, with contributions by Maxine Anastasi, Matthew Hobson, Victoria Leitch, Geoffrey Morley, Nicholas Ray, Candace Rice, Erica Rowan, Ben Russell and Nichole Sheldrick

Introduction

The second season of the Utica project, a collaboration between the Tunisian Institut National du Patrimoine and the University of Oxford, directed by (in alphabetical order) Elizabeth Fentress, Faouzi Ghozzi, Josephine Quinn and Andrew Wilson, took place from 31st August to 29th September 2012, followed by a 4-day geological coring season, jointly with Jean-Philippe Goirans geological team, Hakim Abichou, Imed Ben Jerbania, Ouafa Ben Slimane and Jean-Yves Monchambert, to determine the topography of the ancient shoreline. This report summarises the excavations; the coring will be reported on separately. The project aims to investigate the urban development and economy of Utica through a combination of topographic survey, geophysics, coring, excavation, pottery studies and structural survey. A 2-week pilot season in 2010 laid the groundwork for our excavations in 2012, establishing a new and more accurate plan of the city, and demonstrating the potential of geophysical survey to contribute to understanding city layout and topography.1 In 2012 our project focused on four excavation areas (work in Area 1, investigated in 2010, was not further pursued). Area 2 included a large public building of the Roman period, almost certainly the citys basilica, whose foundations cut through earlier levels, and which was demolished in the late Roman period, with early Islamic occupation overlying part of it. In Area 3, the Maison du Grand Oecus, work concentrated on understanding and exposing part of the range of rooms to the north of the peristyle, partially excavated in the 1950s, with the eventual aim of obtaining information about the nature and dating of the abandonment of the house and in preparation for its conservation and presentation to the public. New areas of excavation were opened up in Area 4, to investigate the urban context of a topographic feature thought possibly to be a rampart, and where geophysical survey in 2010 had suggested there might be a kiln; and in Area 5, a complex of tanks identified on Lzines plan as cisterns, but identified in 2010 as fish-salting vats, with the aim of resolving this question.

N. Kallala, E. Fentress, J. Quinn and A. Wilson (2011). Survey and Excavation at Utica 2010. Available at http://www.academia.edu/1439423/Survey_and_excavation_at_Utica_2010.

Area 2: The Basilica2 Elizabeth Fentress, Faouzi Ghozzi, Benjamin Russell Excavations north of the Decumanus In 2010, excavations north of the road surface running across the middle of Area 2 exposed the remains of a small early Medieval building, Building 2, sitting on top of the remains of a much larger Roman structure, Building 1. This earlier building comprised a wide central navenow represented by the compact mortar surface, north of our trenchwhich was surrounded by a double row of walls, robbed out by two major robber trenches; these trenches measure roughly 2.503 m and 3.50 m wide respectively, and are 33.25 m apart. Between them a mortar preparation surface for paving was discovered. The working hypothesis is that the inner trench was dug for the removal the foundations and stylobate of an interior colonnade of Building 1, while the external one follows the line of the outer wall of the building. The complex clearly extended to the south, however, where another mortar preparation surface was discovered, measuring roughly 4 m northsouth, separated from the road by a narrow robber trench approximately 1.50 m wide. At the moment it is thought that this structure might represent a portico running along the south side of Building 1, which is perhaps the civil basilica of Utica. In 2012, work in this area focused on excavating Building 2 and emptying the three robber trenches associated with the spoliation of Building 1, so as to understand the original form and final destruction of this monumental structure. The earliest levels exposed this season were found in the central robber trench, which cut through some sort of mudbrick structure, with an associated floor located approximately 0.80 m below the floor of Building 1. Further work in this area in 2012 allowed a number of observations to be made about these features. First, it was discovered that the mudbrick layer, in the north section of this sondage is in fact a truncated wall, [2101], at least three courses of which are complete (fig. 1). The mudbricks used vary in size but are on average 0.250.30 m long, 0.15 cm high and wide. This wall is aligned roughly west-northwesteast-southeast, slightly differently from the later wall of Building 1 which is closer to eastwest. The pale lime surface identified in the previous season appears, then, to have been a floor associated with this portion of extant mudbrick wall. This floor surface has not been excavated, but a fragment of black-glazed fine ware found embedded between two mud bricks has been tentatively dated to between the fifth and third century BC. This structure is not the only one revealed in the robber trench. At its eastern end two low walls
The team excavating the site included Elizabeth Fentress, Faouzi Ghozzi, Ben Russell, Ismahen Ben Barka, Besma Huiji, Mouna Hbachi and Julia Nikolaus.
2

Figure 1. Mud brick wall [2101] on the north side of the construction trench for the basilica wall. (EF)

meeting at a right angle were exposed. Constructed of mudbrick or pis on a stone base, these walls were cut by the construction trench of Building 1. Although work in this area is still on-going, initial examination of the northern section revealed the truncated walls and floor surfaces of at least three other earlier structures, destroyed when Building 1 was built. All of these had mudbrick or pis walls, though none were as well-preserved as [2101]. Documenting these remains further will be a project for next season. The earliest features of Building 1 still extant are the preparation layers of the floor surfaces (fig. 2). These are all constructed in a similar way out of mortared rubble. On top of these, degraded mortar surfaces were found. A few fragments of paving slabs, in grey limestone, were found in place, while the imprints of slabs could be seen in the mortar surface. These marks indicate that the paving slabs were on average 11.25 m wide and 1.602.10 m long, slightly smaller than the slabs used for the road surface. Interestingly, though, these two interior surfaces lie at roughly the same height above sea level as the exterior road surface. Deposits found on top of the floors tell us something about the abandonment and destruction of Building 1. The surviving fragments of paving were found beneath large pieces of fallen granite column. It had evidently been too difficult to move these

Figure 2. The three robber trenches of the basilica's southern wall, partially excavated at the end of 2012. (EF)

heavy obstructions to get at the paving below, indicating that the stripping of these slabs occurred only once the building had collapsed. Layers of ashy rubble, containing large quantities of pottery, were found on the stripped mortar surfaces which might represent working surfaces related to the first phase of spoliation of Building 1, probably in the early Medieval period. Although none of the walls of Building 1 survives, the location, width and depth of the robber trenches in this area reveal a lot about the form of the structure and its later destruction. Following the accumulation of the rubble layers, the earliest feature that survives is the robber trench running along the north side of the road. This trench cuts through the rubble layers, the mortar surfaces, and the preparation layer, but as we will see is overlain by Building 2, which is clearly dated to the late 9th century AD. This robber trench follows the line of a wall, running parallel to the road, the foundations of which were built out of limestone ashlar blocks, several of which remain in place. The fills of this robber trench are yellow and sandy and full of architectural fragments, including pieces of pilaster, pilaster capital, cornice and architrave blocks, capitals and wall revetment. The fact that this material was simply discarded and then used as fill shows that in this early phase of spoliation it was large blocks for building rather than raw material for lime production that were being sought out. The dimensions of this robber trench and the ashlar blocks found 5

in it indicate that the removed wall was roughly the same size and form as the front wall of the shops along the road to the east of our excavation area. This lends some support to the theory that this was a portico wall, or perhaps the front wall of a row of small units backing on to the load-bearing south wall of Building 1. Not long after the back-filling of this trench a small structure was erected over it in roughly the middle of our excavation area. This building, Building 2, had eastern and western walls built out of re-used ashlar blocks and granite column fragments, sitting on low pis and rubble foundations. This structure probably backed to the north onto the exterior wall of Building 1, while most of its southern wall, except for one large column fragment, was removed by the cut made by later excavations along the north side of the road. Despite these truncations this building is well preserved (fig. 3). Its first floor was a thin lime plaster surface applied directly on top of the cleared Roman preparation layer. The building was divided in two by a narrow partition wall. To the north of this wall a square column fragment, 0.75 x 0.80 m, was placed in the middle of the room, perhaps to act as a post support or a table. South of this dividing wall a small hearth was excavated. The ceramics from this first phase of habitation date predominantly to the 9th century AD. A second phase of occupation is revealed by a compact light grey layer containing abundant ash deposits.

Figure 3. Building 2 from the south. (EF)

To the west of Building 2, a second structure was revealed, now known as Building 5. The east wall of this building was again erected in a shallow foundation trench, cut through the rubble and mortar layers down to the level of the Roman preparation layer. The interior of this building will be explored further next season, but at least in terms of its construction and level it would appear to be broadly contemporary with Building 2.

Following its abandonment, Building 2 was covered by a succession of deposits, There are no signs of intensive habitation at this level, except for a post hole, just west of Building 2, and a shallow trench. These features could be contemporary with the opening of the large robber trenches to the north. The exact date of these northern robber trenches and the final phase of spoliation of Building 1 remains uncertain. These robber trenches cut through all of the Roman and post-Roman layers exposed to date and are overlain only by topsoil. Their size and systematic character suggests that they could be as late as the early twentieth century, though for the moment none of the finds in their fills provides any exact date. The fact that numerous large column fragments in grey and pink granite have been recovered from the fills of these trenches (fig. 2) suggests that whoever was responsible for digging them was primarily interested in recovering large ashlar blocks, of marble or limestone, for burning into lime, which was sold at the site until the 1960s. Any whole column shafts that survived the destruction of the building are likely to have been removed already by this point for re-use elsewhere. Excavation south of the Decumanus Excavation south of the road began in 2010 and continued in 2012. The top layers were relatively hard to read, due perhaps to a certain amount of exploratory excavation in recent years. The emergence of two medieval buildings, Building 3 on the east side of the trench next to the road and Building 4 on the west, led us to expand the excavated area to the south, east and west in 2012. At this point what was originally taken to be a robber trench on the east side of the trench was revealed, after the expansion of 1.75 metres to the east, as a modern cut, probably a sondage by Lzine. This was emptied, and the rest of Building 3 investigated and removed. To the west, Building 4 was revealed down to floor level, while the first layers of what was interpreted as agricultural soil were removed. The earliest layer reached so far is a beige earth layer, faintly compacted, which may be associated with a building as yet unidentified. This was cut by a silo running under the south section of the trench, 1.80 m in diameter and 0.80 m deep. This was filled with relatively clean soil, containing some medieval pottery and a small quantity of animal bone, largely chicken with some ovicaprids. Next to it a wide, shallow hole, 1.60 m wide and 0.40 m deep, was filled with stone. Over the fill of these two features accumulated a layer of grey soil around 0.30 m thick. Interpreted as a cultivation layer, it may in fact have been brought in for the purpose. This soil was cut by a robber trench just over 3 m wide, running parallel to the road around 0.40 m to the south of it. The trench appears to have robbed a wall with the same orientation, around 1 m from the road. Hard-packed earth from what may have been the foundation trench of this wall is visible in the bottom of the cut. Over it a series of fills built up in the ditch, apparently thrown in from both sides. Several of these are rather ashy in character, indicating that there was still habitation in the area. We do not yet know how far this trench continued to the west; indeed, it has only been observed so far in the section cut by the modern sondage.

A structure was built over the filled robber trench that we have found difficult to interpret, known as Building 3. Of this structure the north, west and south walls survive, while the east wall was apparently removed by the modern excavation. Inside the space created by these walls a thick layer of yellow clay plaster was found, smooth and polished. This was interpreted as a floor surface, on analogy with similar surfaces at Stif and elsewhere. It was partially covered by a second, similar surface on the south side of the structure. What is curious, however, is that neither of these floors showed traces of a hearth, pits, or other signs of occupation, so their interpretation must remain in question. The highest layer in this sequence is perhaps to be interpreted as the collapse of walls in pis, an interpretation suggested by its pale yellow colour. A rather large silo was found outside the northeast corner of the building, 2.35 m long and 1.60 m wide, oval in shape with a rather irregular bottom, at most 0.60 m deep. The earth that filled it was relatively clean and dark yellow, with a few bones and a little pottery.

Figure 4. Area II south of the Roman road, showing ninth century structures. (EF)

To the west of the building a hard yellow surface abutted a rather vague linear feature of piled-up earth and stones that ran north-south across the trench. This seems best interpreted as a property boundary, separating Buildings 3 and 4 or, perhaps, some earlier structures, as it appears to pre-date Building 3. It may have been topped by a hedge, but it was certainly not a wall. To the west of it a wide, shallow pit was dug, apparently cutting it. Measuring 2.90 x 2.60 m, the pit was rounded, and up to 0.50 m deep at the centre. It was backfilled with stones and earth, without any particular concentration of bone or pottery, and its shape does not suggest that of a silo. It may have been a pit dug to recover earth for building 8

pis walls, perhaps those of Building 4. Just to the northwest of this, a second pit had straight sides, measures 2.40 x 1.70 m and is 0.60 m deep. This appears a more plausible silo. Building 4 is found on the extreme western edge of the trench. The north wall was marked by a large orthostat at its northeast corner. It continues as a masonry socle, marked by a door around a metre to the west, which was subsequently blocked. It bonded to the west wall which terminated with a small orthostat, joining it to the south wall. This wall was very poorly preserved, only 0.30 m wide as it disappeared under the west section. The door was probably found in the southeast corner, at least in the second phase, when the north door was blocked. Inside the building the earliest floor was of beaten earth, with a patch of burning that seem to indicate a kanoun, or brasier. A small step, 0.10 m high, separates it from a second room, to the west, with a very hard beaten-earth floor. Over both floors was found a deposit of ash around 0.1 m thick, perhaps indicating the abandonment of the building. A deposit of cultivation earth was found south of the building, and abutting it. Buildings 3 and 4 are not necessarily contemporary. Indeed, if the agricultural soil that predates both the robber trench and Building 3 is equivalent to that which postdates Building 4, Building 4 is earlier than Building 3. However, it appears likely that Building 3 is not the first structure in its immediate area, as the silo to the south of it very probably relates to another, earlier, building. Like Buildings 2 and 5, however, the little structures seem to relate to the road, and to sit on small plots along it. Postholes in the road further to the east seem to suggest a further structure some 10 m away. The silos of both buildings must relate to grain storage, although no seeds have yet been identified from them. The bones are almost exclusively of chicken, indicating a rather lean diet. There is as yet no connection between this little settlement and the Roman or Byzantine structures beneath it. The function of the area south of the road in the Roman period remains a mystery. It is assumed to have been the forum of the city, but neither our excavations nor the numerous undocumented excavation trenches to the south seem to reveal anything resembling a unified pavement over the very early Punic tombs exposed in some of them. We hope to gain a better understanding of this site in the next campaign.

Area 3: The House of the Large Oecus3 Geoffrey Morley and Nichole Sheldrick The Maison du Grand Oecus was partially excavated in 1957 by Alexandre Lzine and Paul Veyne. The excavations uncovered half of the peristyle court with its central pool, and parts of the north and west ranges, but were not completed and remain substantially unpublished. In the 1970s, The Corpus des Mosaques de la Tunisie project, led by Margaret Alexander and Mongi Ennaifer, documented the
The team who excavated the site also included Geoff Morley, Nichole Sheldrick, Walid Ammouri, Yassin Grami and Andrew Wilson.
3

Figure 5. House of the Large Oecus: position of 2012 excavation of the Veyne baulk. (NS)

visible remains, focusing especially on the mosaic floors, and carried out some limited sondages, through which they dated the house to the late 1st or early 2nd century AD.4 Our own work in 2010 explored two sondages and aimed at stabilising the worst of the damage to the mosaic and opus sectile floors that had remained exposed since their initial uncovering. The focus of the 2012 season in Area 3 was the excavation of a range of four rooms in the north part of the house: Rooms XIX, XX, XXI, and XXII. The initial jobs this season were to remove all the vegetation and the immediate topsoil from the surface of the site and to clean and re-establish the limits of the earlier excavations.5 The French excavations had chased and isolated the walls on the north, south, and west sides of this range of rooms, as well as a strip of the pavements at the north and south ends of the rooms, leaving a roughly L-shaped island of intact stratigraphy within them (fig. 5). The eastern limit of excavation was deliberately placed with the aim of creating a half-section of Room XXII. The southern edge of the baulk was cut back as a single context in order to record the section and give us a preview into the stratigraphy of rooms XIX, XXI, and XXII. It also re-exposed the pavements at the south ends of these three rooms, allowing us
Dulire, C. 1974. Corpus des mosaques de Tunisie, vol. I.2. Utique. Les mosaques in situ en dehors des insulae I-II-III. Tunis: Institut National d'Archologie et d'Art. 1-4. 5 Photographs from the Corpus des Mosaques investigations (Dulire 1974: Plates IX (bottom), X, XI, XIII (top)) show that at that time this cut was closer to vertical (or perhaps was made so during that project); by 2012 subsequent erosion had rounded off the edges and created more of a slope.
4

10

to re-assess their current conditions. On the north side, the pavement of Room XIX was also re-exposed during initial cleaning, while those in Rooms XX and XXII were only partially revealed.6 Given the constraints of time and resources, however, none of the four rooms was excavated fully to pavement level this season. Room XIX was paved with a black and white geometric mosaic (No. 160).7 The lowest levels revealed by excavation in the room were a layer of pis collapse and a small deposit of ash located in the southwest corner of the trench. It is hoped that the finds from these deposits might be able to date the final phase of use in this room before they were sealed by the remains of the west wall of the room. This was of opus africanum construction, one of the piers being found amongst the rubble. The existence of two rooms between Rooms XIX and XXII rather than one had been correctly surmised by the authors of the Corpus des Mosaques from the different pavements found on the north and south sides of the unexcavated area. Room XX was paved with a black and white geometric mosaic (No 161),8 while Room XXI was paved in square slate tiles (No. 162).9 Their hypothetical reconstruction of the house therefore placed a dividing wall at what appeared to be the south edge of pavement No. 161 in Room XX, approximately 2.4 m south of the rear wall of the building.10 Our excavations located the western end of this dividing wall at approximately 4.5 m south of the rear wall. It seems probable that mosaic No. 161 therefore only covers the north part of Room XX, and that the south part is either paved in an as yet unknown fashion, or not paved at all, as is the case in Room XII just to the west, perhaps leaving space for a staircase or some type of furniture. This issue will be resolved in the 2013 season. As in Room XIX, the earliest layers above the pavement in Room XXI are currently only visible in the south section (fig. 6). Extending out from the section, just above the level of the pavement are plastered wall fragments from a wall collapse. Above this wall collapse appears to be a thick levelling layer of pis, potentially equal to that in Room XIX, but now physically separated by robber trench 3120 over the wall between the two, of which only the foundations remain. In addition to several fragments of iron cans and other objects, this trench contained multiple small ash deposits, which have been interpreted as possible remains of fires used by workers involved in the robbing. A second trench running eastward from 3120 almost certainly represents the fill of a robber trench for the E-W dividing wall between Rooms XX and XXI discussed above. Various intercutting pits were also excavated this season. These may generally be interpreted as pis quarries, and contained very little material, although one yielded a small fragment of 19th-century glazed ware.

The pavements were not given new context numbers during this season and are referred to by the numbers assigned to them in the Corpus des Mosaques. 7 Dulire 1974: 13-14, Plate IX (bottom). 8 Dulire 1974: 14, Plate X. 9 Dulire 1974: 15, Plate XI (top). 10 As seen in Dulire 1974: Plan 2, Plate X

11

Figure 6. Contexts in the south section of the baulk. (NS)

An opus sectile pavement of alternating giallo antico and grey marble slabs (No. 163) formed the floor of Room XXI,11 above which the section shows thick layers of pis. Several pit features were cut into this level. The earliest appears to have been 3125 an ashy grey pit or trench, rectangular in section. This was cut at an unknown point by a deep, bell-shaped pit, 3146 (fig. 7). Its shape indicates that this pit was a silo, but its fill contained many large stone blocks as well as finely carved marble architectural fragments. The decorated pieces are of the same type recovered from the excavations of Building 1 in Area 2: perhaps they were intended to consolidate the hole after it was emptied. However, they appear to confirm the dating of the activity on this site to the same period as the robbing of the basilica. The site was probably adjacent to the huts for the workers on the robbing enterprise, and further large, rounded pits like 3134 are again to be interpreted as quarries for the construction of the Figure 7. Silo filled with architectural fragments from huts. the basilica. (EF) All of these features were overlaid by a single layer, (3113). The finds recovered from this context range from the 6th century BC to the 20th century AD. It has therefore been interpreted as the spoil from Veynes excavations.

11

Dulire 1974: 15, Plate XI (bottom).

12

Area 4: Rampart and Kilns12 Andrew Wilson, Matthew Hobson and Candace Rice A new area was opened in 2012 to the south-east of the large second-century Roman baths, with the aim of investigating the nature and topographic context of a sharp break of slope, thought by Lzine to be part of a circuit of ramparts enclosing the high ground of the ridge dominating the site. The area excavated included a slot through the putative rampart, and an area excavation in front of it to examine the topographic context of what would be the immediate field of fire in front of the possible fortification; geophysical survey of this area in 2010 had shown that this zone incorporated a large circular geophysical anomaly and several smaller ones. Excavation showed that the depth of topsoil covering archaeological features was very shallow, and that several structures appeared, truncated, directly below topsoil. The anomaly found by the 2010 geophysical survey proved to be a large lime kiln, built into the steep side of an embankment. The removal of topsoil in a more gently sloping 10 x 10 m area to the north confirmed that the smaller anomalies visible on the magnetometry results were also kilns, perhaps used for the production of pottery; a large dump incorporating some pottery production waste and repeated instances of the same ceramic forms was found immediately between them and the lime kiln. These smaller kilns have a complex relationship with several phases of building activity. This area was planned in preparation for removing some of the extensive mud brick or pis demolition deposits at the beginning of the 2013 season. The possible rampart A slot 8 m long and 3 m wide was cut across the line of the sharp break of slope to investigate whether this was a natural topographic feature or part of a man-made fortification, either a Punic or a Late Antique rampart. Excavation here revealed parts of a Roman structure represented by two walls, [4005] and [4006], joining at right-angles, and overlain by weathered mud-brick or pis demolition, and then by topsoil. The building appears to have been truncated by the circular lime kiln downslope to the NW, but the precise sequential relationship remains to be confirmed through further excavation. The structure lies partly across the break of slope and at a 45-degree angle to it. The walls are built of mortared rubble concrete and the northern face of the structure is covered with a whitish hydraulic plaster containing some pink crushed terracotta inclusions. The upper part of [4005] is destroyed at the north-east end but intact towards the south-west, where it is also sealed by the plaster, suggesting that the structure was an open tank, possibly for clay preparation associated with the smaller kilns to the east. A terracotta pipe, exposed just at the south-eastern edge of the trench running against or through wall

12

The team who excavated the site included Andrew Wilson, Matt Hobson, Tyler Franconi, Candace Rice, Maxine Anastasi and Soumaya Trabelsi.

13

Figure 8. Area 4, Early Roman structure defined by walls [4005] and [4006], and lime kiln [4003] under excavation, facing S. (AW)

[4005], may provide some support for this view. Wall [4006] continues southwest beyond the junction with [4005], ending in what appears to be an external buttress made of sandstone blocks; adjacent to this buttress the northern face of [4005] slopes sharply as a steep ramp, again plastered. The walls were abutted to the north by a layer, which underlay the surface that sealed the fill of the construction cut for the kiln. Further work is needed to resolve the question of the so-called rampart, but the discoveries of 2012 seem to exclude the idea that the sharp break of slope in this area might be a Late Antique rampart. This is because the uppermost (and thus latest) structures encountered appear to be of Roman date, associated with the industrial activity of the kilns; they lie partly across the expected line of the rampart; and even if the rampart was a few metres further uphill, they would interfere with a clear field of fire in front of any fortifications here and compromise their defensibility. We cannot at this stage, however, fully exclude the possibility that the break of slope might represent an earlier, perhaps Punic, rampart. More work on this question is required in 2013. The lime kiln [4003] The circular anomaly shown on the geophysics plot proved indeed to be a large circular kiln, 6.2 m in internal diameter, cut into the steep bank thought by Lzine to represent the rampart. The lower part of the kiln was cut into the ground, and built

14

of mud bricks 13 bonded with mud mortar within a backfilled construction cut. The bricks of the walls had been fired in situ by the firing of the kiln, having turned greenish on the inner face, fading to red on the outside. The internal surfaces were covered with a thick deposit of yellowish white lime. Above the ancient ground level the kiln originally rose as a free-standing structure, which had survived better on the eastern side against the slope of the terrain, to the point where the springing of the dome could just be made out. Although the interior was not bottomed in 2012, the internal height measured from the limit of excavation to the top of the standing remains where the dome started to curve over was over 4 m, and the original height with the dome intact would have been substantially more.
Figure 9. The lower lining of the kiln. (AW)

In the lower part of the kiln the internal diameter is reduced, and the lower part of the kiln was lined with roughly squared limestone rubble blocks,14 mortared with a light-whitish grey limestone mortar with small grade aggregate (fig. 9). The somewhat damaged top of this stepped-in lower lining formed a shelf. The stokehole was on the downhill, north-western side, and sited to catch the prevailing north-westerly winds as they hit the bank, creating an updraft with excellent conditions for firing. The rake-out pit in front of it was filled with alternating layers of ash and lime, each possibly representing a firing of the kiln.15 There are signs of subsequent relinings of the kiln: a second lining of the kiln, in mudbrick baked in situ and bonded with a friable light yellow mortar, sat on the stepped-in kiln wall. A separate relining is visible in the northern quadrant, reducing the interior diameter of the kiln and narrowing the usable shelf width. This relining was again in mud bricks, with a fill of earth and stones against the original lining; it too sat on the stone shelf of Phase I. The interior face of both linings, where exposed, was covered in a deposit of white lime. A heat-affected reddish deposit appeared completely to surround the kiln, except for a noticeable break where the rake-out pit was located, and seems to be the fill of the construction cut for the kiln, not yet fully observed in plan. An exploratory slot across the rake-out pit also encountered the construction cut for the kiln, located over a metre away from the kiln wall, and this resulted in the slight overcutting of

13 14

Sample dimensions 0.080.09 m x 0.240.23 m. Sample dimensions: 8 x 1427 x 11 cm. 15 Cf. D. A. Jackson, L. Biek and B. F. Dix (1973), A Roman lime-kiln at Weekley, Northants, Britannia 4: 128140.

15

the rake-out pit at the eastern edge of the slot (fig. 10). This does, however, mean that we recovered some ceramics from the upper fill of this construction cut. More secure dating material may be collected next season, as it is possible that some ceramics from the fills of the rake-out pit 4024 could have become intermixed with those collected under the number (4023), which related to the construction deposits. Nonetheless, the prospects for dating the kiln look good, as the fill of the construction cut is full of large, diagnostic ceramic sherds. The fill of the kiln was halfsectioned and the western half excavated (fig. 10). The lowest fill reached in 2012 was (4018), a loose, mid-pinkish brown, powdery, soft silt, with occasional inclusions of large stones and fragment of mud brick from which the kiln was constructed. It appears to have Figure 10. The rake-out pit, 4024 and construction cuts for developed as a gradual silting up of the kiln chamber along with the kiln. (AW) occasional falling debris of the weathering structure. This deposit implies abandonment rather than any kind of deliberate infilling. The layer was overlain by (4016), still predominantly a silt, with frequent large angular stone inclusions, white powdery degraded mortar, as well as large solid lumps of mortar. In the centre large chunks of the kiln vault were visible, with several integrated courses of mud bricks. This must represent the collapse of the kiln superstructure, and was notably more consolidated than the uppermost fill over it, (4010), a mid-brown, slightly sandy silt, representing the final filling-in of the kiln through a process of gradual silting and colluvial action. It was sealed by topsoil. On the south side of the kiln a deposit very rich in pottery (4015) was identified as distinct from topsoil but is evidently disturbed and not a good sealed context. Up slope, on the eastern side of the kiln, an ashy layer, very rich in ceramics, was observed in the west-facing section of eastern baulk after removal of the topsoil. It appears to be one of a series of thin layers tipping off the mound created by the roof of the kiln. Presumably these layers relate to the use-life of the kiln. The north-eastern part of the trench On the lower-lying ground to the east of the lime kiln, where before excavation two shallow sub-rectangular depressions had been visible, the geophysical survey in 2010 showed several black and white anomalies. Here an area 10 m x 10 m was opened up to investigate these anomalies, that would seem to have interfered with any clear field of fire in front of the so-called rampart. A series of features and deposits here were exposed in plan after the removal of topsoil, but their stratigraphic relationships with each other have not yet been established. They included a dump 16

Figure 11. Half-sectioned fills of lime kiln 4003. (AW)

of ceramic material, a section of terracotta water-pipe, and some remnants of walls associated with a depression, possibly a tank or collapsed cistern, and some small kilns, one of which lay entirely within the trench, while the other two were partly obscured under the northern baulk of the excavation. One of the geophysical anomalies proved after removal of topsoil to be a small circular kiln [4022] which appeared to have truncated earlier features (fig. 13). Immediately west of the kiln the upper part of a masonry feature [4027] was exposed with plaster on its western and northern faces. It has evidently been truncated by the kiln, and may form part of the same wall as a mass of rubble, the top of which was exposed to the east of the kiln. Kiln [4022] is c. 1.70 m in internal diameter, with a wall in mud bricks that had fired greenish on the inside and reddened on the outside. It was surrounded by a distinct ring of burnt earth. Immediately to the south of the kiln, a marked sub-rectangular depression, visible even before excavation. The stratigraphic
Figure 12. Masonry features in the north of the trench. (AW)

17

relationship between these features, the kiln and the remnants of the wall represented by [4027] has not yet been established. Parts of two other kiln linings, as yet unexcavated, in the northern corner, corresponding to other anomalies visible on the geophysics, were visible within the north and north-eastern part of the trench but extended beyond it. In the centre of the trench, a large ashy deposit (4009), very rich in pottery, may represent a dump from the usage of the smaller kilns, assuming that these were indeed pottery kilns. Topsoil directly over the dump was notably rich in pottery and this material was separately collected, as it was suspected of being disturbed material from the deposit below; it included a few wasters of cooking and coarse wares, but there were also a notably large number of the same kinds of forms, all in local (northern Tunisian) fabrics, including amphora rims of Van der Werff 1, and Hayes 191 and 194 cookwares. In association with black-glazed Campanian wares, this begins to suggest local production of cookwares and fish-product amphorae in the first century BC/early first century AD. Uphill from the small kilns, and over much of the central area of the trench, thick deposits of pis demolition clearly derive from a series of stone-footed buildings oriented on the main city grid, as the geophysics indicates and the surviving walls confirm. Two sections of terracotta water pipe run across the centre of the trench directly below topsoil and cutting much of the pis demolition: these are [4048], described above in relation to the possible tank, and [4047], which is represented by a nearly complete length of pipe and the female flange of the next section immediately downhill (fig. 13).

Figure 13. Terracotta pipe [4047]. (AW)

It thus appears that this area was in early Roman times a suburban industrial quarter, on the western edge of the main residential zone (to the north-west, a series of mausolea flanking a road heading west out of town, and the large secondcentury AD baths, built where space was available on the edge of the city, confirm the suburban nature of the area). The apparent co-location of pottery and lime kilns is interesting, and perhaps results from the deliberate exploitation of the topography and wind conditions. Such co-location is paralleled elsewhere, for example at Weekley (Northamptonshire, UK).16

16

D. A. Jackson, L. Biek and B. F. Dix (1973), A Roman lime-kiln at Weekley, Northants, Britannia 4: 128140.

18

Area 5: Excavation of the fish-salting vats17 Erica Rowan and Nicholas Ray In 2010 a series of six tanks was identified by Ouafa Ben Slimane and Andrew Wilson as fish-salting vats (fig. 14). Only the eastern, western and southern walls of the three northern vats (4, 5, 6) have survived and thus these vats do not contain any material due to collapse and erosion. The northern wall of the southern vats (1, 2, 3) has survived to varying heights and these vats contained accumulated material. Figure 14. The fish-salting vats. (EF) During the 2012 season, we carried out a very limited exploratory excavation, cross-sectioning vats 13 in order to confirm the fish-salting hypothesis and to determine the sequence of events that led to their abandonment and subsequent filling. The area within and around the vats was also cleaned and all vegetation removed. Environmental samples were taken from within the vats for sieving and flotation for the recovery of small fish bones and fish scales. The six vats are arranged in two lines of three units (fig. 16) and were all built as a single masonry construction. The walls and floors of the vats were then coated with a plaster lining and a lip at the intersections of floor and wall was then applied. The presence of the lip makes it impossible to ascertain if the wall plaster overlies that of the floor, but this is the most probable interpretation of the sequence. The only vat not to possess this format is vat 5, which has been re-floored with opus spicatum and the walls coated with a coarser plaster. This vat is connected to vat 6 just above floor level by a lead pipe running through the wall dividing these vats. The dimensions of the vats are as follows: NS Dimension (m) Vat 1 Vat 2 Vat 3 Vat 4 Vat 5 Vat 6 Total 2.9 3.0 3.0 2.3 2.2 2.4 EW Dimension (m) 3.9 1.6 5.0 5.0 1.5 4.1 Minimum Depth (m) 2.2 2.2 2.2 2.3 2.2 2.2 Approx. Minimum Volume (m3) 24.9 10.6 33.0 26.5 7.3 21.6 123.8

17

The team excavating the vats included Erica Rowan, Nick Ray, Julia Nikolaus and Soumaya Trabelsi.

19

Figure 15. Schematic plan of the vats.(ER)

Vat 1 is located in the south western corner of the structure and was connected to vat 6 by an opening approximately 0.4 m wide in the central portion of the north wall. Although the actual total depths for each of the vats are unknown, because the original top of the walls is nowhere preserved, the minimum depth of this vat is 2.2 m, giving a minimum volume of approximately 24.9 m3. The eastern wall, which divided vat 1 from vat 2, was badly damaged and only survived to a height of about 0.6 m. The excavation and cleaning of the earliest levels confirmed that the gap between vats 1 and 6 was an intentional connection between the two vats and was part of the original construction. This was verified by the lining lip, which continued seamlessly around the eastern and western corners of this opening. The floor of vat 1 had been damaged and little of the original floor lining survived. The plaster lining of the walls was also badly damaged. Samples were taken from the fill layer just above the floor of the vat but they produced little in the way of fish scales and bones. The damage to the floor in vat 1 indicated that there had been post-abandonment use of this tank. Against the north wall of the vat, east of the connection with vat 6, was a shallow cut into the floor, filled with an ashy deposit, suggesting a shallow fire pit (sheltered by the north wall from wind). The pottery from these floor layers, as well as from those in vat 3, indicate that the vats were re-used sometime between the 9th and 12th centuries AD as temporary shelters, with no other activity being evident. The remaining fill layers do not show any signs of occupation and indicate a gradual accumulation of material over time. The 1520 cm of soil above the floor of vat 1 consisted of a sandy fill with evidence of wall collapse. This suggests that the northern and western walls had begun to collapse shortly after the vats went out of use. Above this deposit, a layer of very loose, almost sterile, windblown sand filled the eastern end of the vat. This context also stretched across vat 2 and the western half of vat 3, indicating that by this time the side walls of vat 2 had collapsed to their present height. Vat 2 is the south-central vat, situated between vats 1 and 3. The northern and southern walls survived to heights of approximately 2.2 m. The late destruction of its 20

eastern and western walls, in combination with its small size, meant that vat 2 had a different depositional history from vats 1 and 3. The floor plaster of vat 2 was very well preserved and did not contain any cuts as in vats 1 and 3. This suggests that this vat did not undergo the same post-abandonment activity seen in the larger vats, probably due to its small size. The plaster lip lining was almost entirely intact and the wall plaster was preserved up to the height of the eastern and western wall collapses. Unlike the floor cuts in vats 1 and 3, the well preserved floor in vat 2 suggests that it was left abandoned after the fish salting tanks went out of use. On-site sieving during the excavation of this tank produced fish scales and so the layer immediately above the floor was sampled extensively for fish scales and fish bones. Within the southern baulk of vat 2 was a whitish-grey mortar collapse with a brown organic layer beneath. The location of the mortar suggests that it comes from the southern wall of the vat. The brown organic layer extends to the floor of the vat which means that it was there prior to the intrusion of the windblown sand. It is thus likely that the excavation of the southern portion of the vat will provide more information about the chronology of the collapse of the southern wall of the vats. Vat 3 is located in the south-eastern corner of the structure and is similar in form to vat 1, complete with a connecting channel (approx. 0.4 m wide) to vat 4 in the central portion of the north wall. It is not possible to tell whether this connection continued to the full height of the walls, or whether there was a ceiling to it, because the wall at this point reached no higher than 0.4 m (the same was the case for vat 1). As with the other vats, the actual total depth for this vat is unknown, although the minimum depth of 2.2 m provides a minimum volume of approximately 33 m3, making it the largest of the 6 vats. The total capacity of this complex would have been in excess of 120 m3. Despite systematic sieving and environmental sampling, little was recovered in the way of fish bones and scales, even at floor level. This was most likely a result of prolonged exposure to the elements once the vats no longer functionally served for fish salting, combined with post-abandonment occupation. This occupation was most apparent in the eastern end of the vat, with several cuts into the floor. These appear to have served two functions: a fire pit, suggested by the ashy deposit contained within it, and a circular cut that did not contain an ashy fill, suggesting that it was designed to support a jar or similar vessel. One further, roughly rectangular cut was found in the floor, this time within the channel connecting vats 3 and 4. The ash in the fill again suggests that it was intended for a hearth. An interesting post-abandonment feature is a probable game cut into the floor in the west of the vat, comprising three rows of approximately ten to twelve small divots (the exact number is uncertain because of the degraded state of the plaster surface). Over this a sequence of deposits accumulated, again suggesting abandonment and windblown material. The early levels again provided ceramics dating to the 9th12th centuries within 10 cm of the floor level. 21

Pottery Report Maxine Anastasi and Victoria Leitch General aims and methods The 2012 September season at Utica aimed to set up a recording system and methodology for the following seasons, and to record all the pottery from the 2010 season and as much material from the 2012 excavations as time permitted. The pottery was recorded in an Excel spreadsheet. In addition, drawings, marked as Featured Vessels with individual numbers, were done in order to have a record of the most datable and common forms, and to start to build a typology of the unknown forms, mainly local Roman and Islamic-period coarse wares. A programme of research into the fabrics has also been undertaken: small chips were taken from each of the Featured Vessels, and photographs taken to show the surface treatments. The plan is to match the samples to existing information on fabrics, such as Peacocks Carthage fabric series, and also to start to isolate the local fabrics, with particular reference to the forms that came from the kiln site in Area 4. A catalogue of drawings with their fabric chips attached has been created, and work comparing fabric and forms, and starting an Utica typology will be carried out ahead of the next season. Such preparation will considerably speed up the work on site in future and give us a better idea about the volume of local pottery in the area and the diversity of imported forms and fabrics over time. Area 2: pottery and interpretation This large area had seen major robbing in antiquity and later, and was thus full of contaminated contexts, with material stretching from the Punic to Islamic periods. An examination of the pottery was useful, however, for informing us about the general chronology of the area and the types of pottery being circulated. The contexts linked to the Islamic houses revealed some known types dating from the 8th to 12th centuries, such as yellow and brown glazed pots and strainer jugs (Figs 16.1 and 16.2). Also, many ribbed amphorae sherds with a similar rim and neck (DAngelo B1/2Fig. 16.4), which are common from the 10th to 12th centuries, plus jug forms (DAngelo E1/2) of a similar date (Fig. 16.3). Interesting to note is the absence of late Roman ARS, suggesting a possible hiatus in the occupation of the site, which future investigations may confirm. Also of interest is the dating of the early structures before the basilica, with a find of a sherd with black gloss, which is linked to the early occupation, somewhere between the fifth and the third century BC.

22

Figure 16.1. Strainer jug, Area 2, 2013. 16.2. Fragment from a strainer jug, Area 2, 2214, possibly th th 10 - to 12 -century or earlier. 16.3 D'Angelo E1/2,Islamic jar, Area 1, 1004, similar to examples from Area 2. 16.4. D'Angelo B1/B2 Islamic jar, Area 2, 2037. 16.5. Consp. B3.1219, with stamp 'ACPE' in planta pedis with barred toes, OCK stamp 2585.149 vessel no. 25724, provenance Pisa or Arezzo. Area 3, 3041.

An examination of the tiles from the basilica deposits was carried out, using weights for the 12 different fabrics examined. Of these, three can be recognized macroscopically as Italian in origin and account for 58% of the total. A further three types are probably also Italian, accounting for another 18% of the total, though scientific analyses are required to further clarify these numbers.
Figure 17. Tyler Franconi sorting tile fragments.

23

Area 3: pottery and interpretation The pottery in Area 3 was all from contexts contaminated by recent archaeological investigations, with material in the same contexts ranging from 600 BC to the present. All the pottery from the 2010 excavations and most but not all of the pottery from the 2012 season has been studied. Despite the mixed contexts, the pottery from this area was useful for showing the range of material in use around the site in general and thus its chronological history. Of special interest are several instances of Van der Werff 1 amphorae, which may have been produced around Area 4. The earliest pottery consists of some Campanian wares (2501 BC), Punic red slip bowls and an askos jug (600400 BC); then from the Roman period Italian Sigillata (40 BCAD 100, Fig. 16.5), African Red Slip ware from the beginnings of production in the mid 1st century AD up to the 5th century (Hayes 91, AD 450500), but with no late forms, which is also true of the lamps, with no examples of the typical large African Christian lamps of the later 5th century onwards. We then jump to early glazed pottery, though very little from the 8th to 10th centuries, but more numerous Islamic amphorae forms of the 10th to 12th centuries, as well as strainer jugs and bifid rim bowls. For these we do not yet have accurate dates, but they certainly appear in North Africa from the 10th century if not before. There is also an absence of local handmade pottery, with seemingly more examples of handmade wares from Pantelleria. Area 4: pottery and interpretation At this kiln site, the insecure surface-level contexts, despite being contaminated, produced pottery datable from a fairly limited timespan, so were in fact useful for giving us an early idea about both the possible products of the kilns and the dating of this important site. The pottery generally suggests use in the Republican to early Roman imperial periods: there is a strong presence of Campanian black gloss pottery of early Republican date (Fig 18.1), while amphorae and ARS suggest that the area was in use up to the 2nd century AD. Two amphorae sherds, an Africana IID and a IIIC place the latest topsoil disturbances in the area to the beginning of the 4th century. Context 4023, into which the lime kiln was constructed, is composed of generally Late Republican-period material (Adriatic thick-walled and Campanian black sand amphorae), with one Hayes 185 lid, no African cooking wares and a handful of Republican fine wares (ITS, ESA and Campanian Black Slip). What is apparent is that there are many more mid-Punic-period forms with their distinct red painted decoration in this context than in the others seen. The only anomaly in this layer is a simple bowl which is coated with a crazed green slip or glaze (UT 994). Further comparisons into this forms identification need to be drawn before more can be concluded. The three contexts (from stratigraphically earliest to latest: 4018, 4016 and 4010) which make up the fill of the lime kiln (4003) are composed of a similar range of forms to the topsoil layers and date to the Late Punic and mid-Roman period; however, the appearance of two possible Islamic jar rims (UT982 and 983) in 4018 24

and body sherds belonging to a LRA2 in 4016 suggest that we may be dealing with a case of reverse stratigraphy whereby the lime kiln was partially filled during or after the Islamic period with material taken from a series of stratified refuse dumps close by. An ashy dump (4009) near the kilns yielded coarseware wasters, presumably fired in a nearby kiln, including a probable jug handle, which, though difficult to date, was found with other pottery of the early Republican period. Additional significant evidence is a rim of a Hayes 194 casserole found in the lining of the lime kiln [4004]; a terminus post quem of around the mid-first century AD can thus be given for this kiln. In general the pottery at this site was surprisingly homogenous, with a very limited number of form types, which points towards production, probably in kilns yet to be uncovered. The common forms included Hayes 191 casseroles (1st century BC to 1st century AD, Fig. 18.2), and Hayes 194 casseroles (Fig. 18.3), a later development of the 191 (though no examples of the Hayes 19, the slipped version of Hayes 194, were found), up until now dated from the mid to late 1st century ADthough if they were being produced in this kiln it might be possible to push this date back. In addition, many variants of the Van der Werff 1 amphorae forms were found, dated from 200 BC to AD 1 (Fig. 18.4), opening up the possibility of amphorae production in the area, which fits in with recent research that suggests amphorae were largely produced and fired alongside cookwares. Some amphorae body sherds also had a very green fabric and outer surface, characteristic of underfired pottery. In addition, a number of unguentaria were uncovered near the small kiln, and next season, when this kiln is excavated, it is hoped that it can be confirmed whether or not they were fired there (Figs 18.5, 18.6). The pottery confirmed that ceramic production was carried out in this area with wasters and under- and over-fired coarse wares, probably in the Republican period, and up to the 2nd century AD. It remains to investigate the small circular kiln and other kiln-like features or anomalies in the area to confirm, if possible, what sort of pottery was produced. Further geophysics work on the surrounding slopes may, it is hoped, indicate whether or not there are more kilns that could be linked to the cooking wares and amphorae finds.

25

Figure 18. 1. Campanian ware bowl, with detail of the foot, 2nd century BC, similar to Morel 2132, Cales, 3rdmid-2nd century BC' from Area 4, 4000. 18.2 Hayes 191 casserole, Area 4, 4008. 18.3. Early Hayes 194 casserole, Area 4 4008. 18.4. Van der Werff 1 amphora, Area 4, 4008. 18.5. Unguentaria, Area 4, 4008. 18.6. Unguentaria, Area 4, 4008.

26

Area 5: pottery and interpretation The fills of the vats contained a mixture of pottery, ranging from Punic amphorae to Islamic strainer jugs; the Roman material included Campanian A and C wares, ITS, ARS; early Roman lamps, and sherds of Dressel 1 and Dressel 2/4 amphorae. Much of this material is clearly residual. The material used in the opus signinum lining of the walls of the vat complex, mainly crushed amphorae, included sherds of Roman amphorae, significantly some with volcanic inclusions, probably from Campania. Imports from this region were most common between 150 BC and 150 AD. Thus, a best guess for the dating of the structure at this point is probably around the 1st century BC to the 1st century AD. Summary and Future Aims The pottery in Areas 2 and 3 demonstrates a complex history of robbing, destruction and rebuilding, reflected in the very mixed contexts, many of which contained material from the Punic to Islamic periods. Notable in these areas in particular is the seeming rarity of late Roman forms, with no datable material between the 6th and the 8th centuries. Area 5 is more straightforward and suggests an early Roman date for the structure, with evidence of occupation in that zone more generally into the Islamic period. The pottery from Area 4, which happily seems not to have been contaminated with later material from robbing or subsequent activity, suggests that this production zone was principally occupied in the Republican period and up to the 2nd century AD. Future excavation might, it is hoped, reveal sealed contexts and dumps that could allow the dating and identification of the products being fired in this area, and which kilns these dumps are associated with. With very few excavated pottery production sites in North Africa, and even fewer well excavated and published examples, this site is of huge importance to the question of the organisation of ceramic and lime production in the Roman period, and in this case also the link to previous Punic traditions.

Acknowledgments We are grateful to the Institut National du Patrimoine and in particular to its Director, Adnan Louhichi, and the director of the site of Utica, Imed Ben Jerbania, for their support of this project. We owe a great debt of gratitude to our sponsor, Baron Lorne Thyssen. At Utica, Naceur Soltani and Hedi Silini make our lives and the excavation easy, for which we are profoundly grateful. The project is directed by (in alphabetical order) Elizabeth Fentress, Faouzi Ghozzi (INP), Josephine Quinn (University of Oxford), and Andrew Wilson (University of Oxford). The image on the first page relates to the Tunisian-French-British coring season that took place in October 2012, and will be reported on elsewhere. 27

28

Você também pode gostar